Bloomberg BW ran an article July, 2018 saying B of A said aggregate QT globally from all central banks will be 4% over next 2 years (2% a year). My opinion: that’s almost nothing if someone adjusts for 2% annual inflation, or perhaps it is a 2% real shrinkage of debt over 2 years assuming a 1% global Developed country inflation rate since inflation is low in the EU and Japan.
Assuming much of the debt has an intermediate term maturity or a due on sale (of  the collateral real estate) clause for mortgages, then the debt will experience portfolio runoff through natural paydowns of principal by borrowers. If Fed does nothing the portfolio will naturally shrink gradually. The Fed estimates a $50Bil monthly runoff rate; that’s $600Bil divided by $4Tril of assets = 15% a year  in 2020 sell off or paydown of debt. Perhaps the absence of a previous policy of preventing runoff means less support for bond prices which means a neutral no-sale policy could slightly increase rates by 15% a year times the 0.5%
aggregate QE induced rate change estimate, which would be 7.5 bps rate increase a year. (I am assuming QE from 2009-2014 only cut rates by 0.5% and the reason rates dropped more was mainly due to the shock of slow global growth and low inflation).

    The Fed’s QT program intends to have the portfolio shrink by $450Bil in 2018 federal FY (which is mainly in calendar year 2019), then in the following year (calendar year 2020) sell $600B, so that’s about a 10% reduction per year in calendar year 2019 and 15% in 2020.
If QE only cut rates 0.5% then a one-tenth annual reduction of balance sheet assets from the QT program would be a mere 5 bps a year rate increase in 2019. However, a sudden global recession would be more powerful and act to steeply decrease rates.
So far in 12 months of QT the assets have decreased by 6%, which is roughly what would occur from portfolio runoff. The Fed stopped QE in October, 2014 and rates went to new lows in mid-2016, partly because of QE in other countries. Rates went up in 2018 because the December, 2017 tax cuts acted to create a short-term sugar high stimulus which increased economic activity and inflation. That increase is now reverting back to the low growth rate of previous years. The crashing US stock market, already a bear market in much of the world, and a bear market for the Fangs stocks, and ex-US the global yield curve is inverted, are in the aggregate, showing that the economy is heading into a cooling down phase. The EU and Japan are unable to get out of their disinflationary low growth mode.
Thus I expect US rates to be slightly below the median of recent years, implying the ten year Treasury could be at 2%; to this one could add a token 5 or 7 basis points for QT each year, so instead of a 2.0% yield it could be 2.07%.
Basically QE 2 and QE 3 didn’t affect the economy that much. The funds simply went into the hands of a few wealthy investors who didn’t spend it because they already had too much wealth. Instead, these surplus funds, in the hands of former bond owners who sold their bonds to the Fed for newly printed money, encouraged a speculative bubble rather instead of promoting the purchase of capital goods needed to expand factory employment. Thus a very slow reversal of the low-impact QE programs shouldn’t make a dramatic change to the economy.
A far more important contingent risk to bond prices is from sudden economic growth including sudden growth in real wages for those in the median income brackets (not simply a bogus mean figure triggered by the elites getting all of the raises.). This is very unlikely since there is so much unemployment in the EU and so many low wage EM countries eager to create low paying jobs by exporting cheap goods. Those countries’ actions can act to transmit disinflation into the US. It has been a long economic cycle of almost 10 years since the last recession. This implies we are overdue for a traditional recession.

   Investors need independent financial advice about the risks of misunderstanding QT.